[LINK] OT: The Quality of Reporting on "COVID-19-Linked" Deaths

Roger Clarke Roger.Clarke at xamax.com.au
Fri Apr 3 11:49:04 AEDT 2020


On 3/4/20 10:41 am, jwhit at internode.on.net wrote:
> Here's a Fact Check article that also explores how the data is
> presented and compared.League tables are always misleading or at least
> can easily be misinterpreted.
> https://www.abc.net.au/news/2020-04-03/fact-file-are-we-turning-coronavirus-corner-flatten-the-curv/12113410

Thanks Jan.  But unfortunately the article's very muddy on deaths and 
death-rates (in the section that starts 40% of the way down the page).

The graph's use of colour, and the inclusion of a legend that has an 
unclear relationship with the lines, makes it almost unreadable.

And this buys into one of the most silly aspects of the entire 
pseudo-statistical mess:
 >There is a second type of death rate. This is sometimes referred to as 
the "case fatality rate".
 >This tells us the number of people who are dying as a proportion of 
confirmed cases.

Let's leave aside the issue of uncertain and probably variable 
test-reliability (as measured by false-positive and false-negative 
proportions).

Countries have adopted very different approaches to testing, and have 
changed their approach over time and space, sometimes frequently.

In most cases, the sample of the population that is being tested at any 
given time is intentionally not random, but targeted.

But the basis of the targeting (the sampling frame, and the manner in 
which the sample is selected from the sampling frame) is highly 
variable, and the execution if it is challenging and highly error-prone.

One result is that within-country counts aren't comparable over even 
short periods, let alone the whole 4-8 weeks to date.

A second result is that inter-country comparisons are completely 
meaningless, because the confounding variables dominate the data.

The article's right to say that "the case fatality rate is not the same 
as an actual mortality rate", and "countries with limited levels of 
testing might appear to have higher mortality rates — particularly if 
that limited testing is being restricted to those patients with more 
severe symptoms".

But it fails to say what matters, which is that "the case fatality rate 
is pseudo-data, shouldn't be reported, should certainly not be compared 
between countries, and is useless as a basis for any kind of 
decision-making".

What's needed is clarity about causes-of-death, and about what that 
tells us can be done for those in danger.

Availability of breathing support?  Probably above all, availability of 
quality medical and nursing staff in appropriate hospital facilities.

But maybe also a severe hosing-down of the alarmism projected by 
politicians, some health policy people, and the media.


> ----- Original Message -----
> From: "Roger Clarke"
> To:"link"
> Cc:
> Sent:Fri, 3 Apr 2020 10:01:35 +1100
> Subject:[LINK] OT: The Quality of Reporting on "COVID-19-Linked"
> Deaths
> 
>   Ruminations on a Friday morning ...
> 
>   The sports results and tables have been replaced by coronavirus (CV)
>   infection-counts and death-counts. And the media declares raw
> numbers,
>   without providing any context to them.
> 
>   This morning's ABC News says that yesterday's 'CV{-linked}'
> death-toll was:
> 
>   Italy 760
>   UK 559
>   Spain 800
> 
>   To get some perspective, that needs moderation by two key variables:
> the
>   countries' poulations and their normal death-rates.
> 
>   Death-rates are quoted as number per thousand of population p.a.
> 
>   So Normal Deaths per Day = (Population/1000 * Death-Rate p.a.) / 365
> 
>   I haven't been able to quickly locate indicators of the degree of
>   variability of deaths per day around the averages shown above, but
> there
>   could be wide variability. In particular, winter in some countries is
> 
>   likely to have higher rates than less-cold times of year.
> 
>   It's not possible with current information to relate CV-caused deaths
> to
>   normal death-rates. As a proxy measure, I've shown below the ratio of
> 
>   deaths yesterday compared with average daily deaths, as a percentage:
> 
>   Country Population Death-Rate Deaths per Day CV Deaths Y'day %age
> 
>   Spain 46m 91 1146 800 69
>   Italy 60m 10.4 1709 760 44
>   UK 67m 9.4 1725 559 32
> 
>   A number of potentially important factors muddy the water:
> 
>   1. Generally, reports fail to distinguish:
>   a. deaths where CV appears to be the only significant factor
>   b. deaths where CV was a significant factor, although not the only
> one
>   c. deaths where CV may have been a factor (e.g. diagnosed with the
>   virus, but nature of death not consistent with CV-caused deaths)
>   d. deaths where CV was present but unlikely to have been a factor
> 
>   The term 'excess deaths' or 'excess mortality' indicates a+b. In
>   German, the word is 'Ueberstirblichkeit', as per:
>   https://swprs.files.wordpress.com/2020/04/mortalitc3a4t-schweiz.png
> 
>   This suggests that Switzerland is experiencing a 'normal'
>   late-winter-flu peak in deaths among over-65s.
> 
>   It may be that there is a great deal of over-reporting due to the
>   inclusion of c. and d. in the numbers appearing in the media. Quoting
> 
>   https://swprs.org/a-swiss-doctor-on-covid-19/, "[It may be that] all
>   test-positive deaths are assumed to be additional deaths".
> 
>   2. It may be that a 'fear-of-the-virus' anxiety factor has
> exacerbated
>   death rates, and even resulted in deaths of individuals who are not
>   infected. For example, populations in countries that are less prone
> to
>   hysteria, such as Germanic northern Europe, evidence very low rates
> in
>   comparison with warm-blooded, Mediterranean countries.
> 
>   3. A variety of reports suggest a very large proportion of deaths has
> 
>   been, throughout, among those over 70 (90%), and a large proportion
> had
>   prior conditions that were life-threatening or could readily become
>   life-threatening (80%).
> 
>   But, apart from a number of specific instances (Wuhan, Iran?, the
>   upper-mid Po Valley, parts of Spain, UK, US), it appears that even
>   deaths among the over-70s may be within the normal statistical range.
> 
>   4. It appears that in both Italy and Spain, many hospitals and
>   aged-care facilities lost a large proportion of their staff, in many
>   cases early in the epidemic. That's because staff from Eastern
> European
>   countries were terrified by panic-ridden reporting and fled home, and
> 
>   large numbers of local staff tested positive and were isolated at
> home.
>   This may have resulted in many saveable patients going untreated and
>   becoming casualties of the epidemic.
> 
>   --
>   Roger Clarke mailto:Roger.Clarke at xamax.com.au
>   T: +61 2 6288 6916 http://www.xamax.com.au http://www.rogerclarke.com
> 
>   Xamax Consultancy Pty Ltd 78 Sidaway St, Chapman ACT 2611 AUSTRALIA
>   Visiting Professor in the Faculty of Law University of N.S.W.
>   Visiting Professor in Computer Science Australian National University
>   _______________________________________________
>   Link mailing list
>   Link at mailman.anu.edu.au
>   http://mailman.anu.edu.au/mailman/listinfo/link
> 
> _______________________________________________
> Link mailing list
> Link at mailman.anu.edu.au
> http://mailman.anu.edu.au/mailman/listinfo/link
> 


-- 
Roger Clarke                            mailto:Roger.Clarke at xamax.com.au
T: +61 2 6288 6916   http://www.xamax.com.au  http://www.rogerclarke.com

Xamax Consultancy Pty Ltd      78 Sidaway St, Chapman ACT 2611 AUSTRALIA 

Visiting Professor in the Faculty of Law            University of N.S.W.
Visiting Professor in Computer Science    Australian National University



More information about the Link mailing list